CARDIOVASCULAR JOURNAL OF AFRICA • Vol 23, No 8, September 2012
448
AFRICA
Mean C-reactive protein (CRP) level was 24.65
±
16.64
mg/l
and white blood cell count (WBC) was 11.11
±
4.12 /
mm³. As
shown in Table 1, none of the other parameters had significant
differences between patients with and without PTE.
D-dimer, fibrinogen and DDFR
The mean values of D-dimer and fibrinogen levels were
3.99
±
3.19
µ
g/ml and 571.4
±
196.1
mg/dl, respectively.
Mean DDFR was 0.712
±
0.643
×
10
-3
.
As shown in Table 2,
D-dimer and DDFR were significantly different between the
PTE and non-PTE groups, but the fibrinogen level did not differ
significantly.
In order to find the best cut-off points for D-dimer,
fibrinogen and DDFR as diagnostic tests for PTE, a ROC
analysis was performed. The ROC curve is used to calculate
the area under the curve (AUC) as a measure of the diagnostic
accuracy. Based on this analysis, for DDFR (AUC
=
0.713,
p
=
0.003)
a value of 0.105
×
10
-3
was 100% sensitive and
1.621
×
10
-3
was 100% specific for a diagnosis of PTE.
Furthermore, this analysis on D-dimer values (AUC
=
0.721,
p
=
0.002)
showed a 100% sensitivity for 0.43
µ
g/ml and a
100%
specificity for 11.5
µ
g/ml for the diagnosis of PTE.
The same analysis on fibrinogen did not show any significant
cut-off point (AUC
=
0.410,
p
=
0.198) (
Table 3, Fig. 1).
Multiple logistic regression analysis was used to create a
model to predict the risk for PTE using the significantly different
independent variable studied at the level of
p
<
0.1.
DDFR,
gender, temperature and WBC were included in backward
stepwise regression analysis. The first model chosen was the
best statistically (
p
=
0.029,
R
²
=
0.577).
The results are shown
in Table 4.
Discussion
No significant difference was found in arterial blood gas and
complete blood count analysis between the hospitalised PTE
and non-PTE patients. These results confirm other studies that
showed arterial blood gas or its combination with other data
could not be used to detect PTE, and if used alone, may lower
the sensitivity and specificity.
20,21
Fibrinogen (factor I) is a soluble plasma glycoprotein,
synthesised by the liver and converted by thrombin into fibrin
during blood coagulation.
22
Fibrinogen may increase in acute
and chronic conditions as an acute-phase reactant with the
same signs and symptoms as PTE. There are few studies on
the relationship between fibrinogen level and PTE. A study by
Palla
et al.
showed that fibrinogen levels in PTE patients (498
±
369
mg/dl) were similar to those without PTE (520
±
268
mg/
dl) (
p
=
0.29.23).
These results are similar to our results, which
demonstrated no significant difference between fibrinogen levels
in the PTE and non-PTE patients. However, Kucher
et al
.
showed
that fibrinogen levels were significantly lower in patients with
PTE (
p
<
0.0001).
9
There is controversy about using fibrinogen levels as a
reliable diagnostic test and it should not be used alone in order to
diagnose PTE. In addition, the fibrinogen level is unpredictable.
It can rise due to acute phases in the ICU or decrease in liver
congestion due to right ventricular failure.
In recent years, several studies have been conducted to
evaluate the accuracy of a diagnostic test to detect acute PTE
in emergency settings. Most demonstrated that D-dimer tests
with cut-off points near 0.5
µ
g/ml could be used as an exclusion
test.
9,23-25
As our study showed, D-dimer levels less than 0.43
µ
g/ml had a 100% sensitivity and a negative predictive value
for ruling out PTE. This confirms data from various studies
demonstrating the ability of D-dimer to rule out PTE.
The only study that evaluated the D-dimer:fibrinogen ratio
in PTE was conducted by Kucher
et al
.
9
They found that a ratio
above 1.04
×
10
-3
had 100% specificity and 57.6% sensitivity
for PTE and a two-fold diagnostic rate compared to D-dimer
alone, with a cut-off point of 7
µ
g/ml with 100% specificity and
29.4%
sensitivity (57.6% vs 29.4%).
9
A recent study however
contradicted Kucher and co-workers’ results.
Calvo-Romero’s study did not reveal a lower fibrinogen
level in PTE patients with a positive D-dimer level, although it
TABLE 2. CONCENTRATION OF D-DIMER, FIBRINOGEN
AND DDFR IN PATIENTSWITHANDWITHOUT PTE; BOLD
P
-
VALUESARE SIGNIFICANT
D-dimer
(
µ
g/ml)
Fibrinogen
(
mg/dl)
DDFR
#
(
×
10
-3
)
PTE positive
4.65
±
3.46 536.73
±
186.32 9.13
±
7.16
PTE negative
2.25
±
2.55 586.33
±
211.06 4.83
±
4.40
p
*
0.006
0.298
0.003
*
From independent samples
t
-
test.
#
D-dimer:fibrinogen ratio.
TABLE 3. RESULTS FROM ROCANALYSIS; BOLD NUMBERS REPRESENT THE
BESTVALUE BETWEENALL CUT-OFF POINTS CALCULATED FROMTHE ROC CURVE
Sensitivity
(%)
Specificity
(%)
Positive
predictive
value (%)
Negative
predictive
value (%)
Positive
likelihood
ratio
Negative
likelihood
ratio
Accuracy
$
(%)
D-dimer (
µ
g/ml) Best sensitivity (0.43)
100
6.9
54.1
100
1.07
-
55.7
Best accuracy (2.43)
70.3
70.1
72.2
67.6
2.35
0.42
70
Previously used
#
(7.0)
24.1
91.9
75
51.9
2.97
0.82
55.7
Best specificity (11.5)
5.4
100
100
51.4
-
0.84
52.8
DDFR* (
×
10
-3
)
Best sensitivity (.105)
100
22.2
58.73
100
1.2
-
62.8
Best accuracy (.233)
91.9
40.4
62.9
81.2
1.54
0.2
67.1
Previously used
#
(1.0)
35.1
84.5
72.2
53.8
2.26
0.76
58.5
Best specificity (1.32)
18.9
100
100
51.6
-
0.81
57.1
*
D-dimer:fibrinogen ratio.
#
The nearest cut-off points in our ROC analysis to the cut-off points used before in other articles and medical references.
$
Calculated by: Accuracy (%) =
true positive + true negative
_____________________
total patients
×
100
